Four separate workforce studies published in the past six months point at the same mechanism: employers told people to use AI without telling them what it was for, and workers responded by performing adoption rather than practicing it. The numbers are no longer a rounding error. They describe a coping strategy.
New research from workforce analytics vendor Visier finds that 48% of employees have exaggerated their AI usage or expertise to colleagues or leadership at least some of the time, and 45% say they feel pressure to use AI despite lacking confidence in how to use it effectively. Visier’s survey of 1,000 full-time U.S. employees, run with qualitative interviews across healthcare, finance, HR and software roles, gives the phenomenon a name: performative AI.
That single statistic would be easy to dismiss as workplace theater with low stakes. It is not isolated. It sits inside a wider pattern that Gallup, Culture Amp and IBM have each measured independently this year, and the pattern is consistent: adoption is climbing, confidence in leadership’s plan is not, and the gap between the two is where performative behavior grows.
None of these four organizations set out to study the same thing. Visier surveyed individual workers about their own coping behavior. Gallup tracks national adoption trends quarterly. Culture Amp benchmarks engagement and culture across a client base of employers. IBM surveyed CHROs and employees separately to compare what leadership expects against what the workforce believes. Four different methodologies, four different sample frames, and they converge on one shape: a widening space between how fast individuals are absorbing AI into their daily work and how slowly organizations are explaining what that absorption is supposed to accomplish.
The adoption curve is real. The guidance curve is not keeping up
Gallup’s quarterly workforce study, fielded in February 2026 across 23,717 U.S. employees, found that 50% of employed American adults now use AI in their role at least a few times a year, up from 46% the previous quarter. Yet only around 10% of employees strongly agree that AI has fundamentally transformed how work gets done inside their organization, and just 41% say their employer has meaningfully integrated the technology at all. Adoption at the individual level is outrunning redesign at the organizational level, and Gallup’s data shows the disruption landing unevenly: workers in service and administrative roles report weaker gains than those in leadership and technical positions, even as AI-adopting companies with 10,000 or more employees post the highest workforce-reduction rates in the sample, 33%, of any size band Gallup tracked. Healthcare and technical workers report the earliest adoption success; productivity gains overall concentrate on individual task completion rather than any organization-wide redesign of how work flows.
That last point matters more than it looks. If the gains are landing at the task level and not the system level, it means most employers have handed out a faster tool without changing the job around it, the org chart, the performance criteria, the definition of a good day’s work. The tool moved. The structure holding the job in place did not.
Visier’s numbers describe what happens inside that gap. In its survey, 54% of employees say their roles have changed significantly because of AI in the past two years, and 51% report significant or moderate change to their actual day-to-day workflow, yet 32% say their employer has no clear plan for how AI affects jobs and a further 27% are not sure whether a plan exists at all. Only 28% believe leadership fully understands how employees are actually using the technology day to day. Put those figures together and the shape of the problem is plain: workers are being asked to change how they work faster than anyone is explaining why. Seventy percent say they are worried about what that means for their career, and within that group, 24% name job loss outright as their primary concern rather than a vaguer worry about skills or relevance.
Culture Amp: the workforce has done its part. Leadership has not done its own
Culture Amp’s first AI at Work benchmark, drawn from roughly 112,000 employees across 123 organizations, put a similar dynamic in harder numbers: 85% of employees say their organization actively encourages AI experimentation, but 42% say leaders have never clearly explained how AI is supposed to help the company reach its goals. Seventy-one percent of employees report that AI makes them feel more productive, rising to 93% among the heaviest users, which is the kind of number that would normally read as an unambiguous win.
Amy Lavoie, VP of People Science at Culture Amp, argues that reading it as a win misses what the data is actually saying. “The workforce is telling us something important. Employees are telling us they understand the risks, they are using the tools, and they feel more productive. That is not a workforce resisting change. That is a workforce that has done its part and is waiting on leadership to do theirs,” Lavoie said in the report.
Lavoie’s diagnosis of why leaders go quiet lines up directly with what Visier’s interview subjects described as the root of performative behavior: nobody wants to commit to a story about the future of work that might be wrong in six months. “Leaders are being asked to explain a future none of us can see clearly yet. We do not know with certainty which roles AI reshapes, which skills hold their value, or what a career path looks like three years out. As a result, leaders may go quiet, because saying nothing feels safer than saying something that turns out to be wrong. The most useful thing a leader can do right now is admit the plan is unfinished, and then keep talking anyway. Certainty is not what employees need from you right now. Candor is,” Lavoie said.
IBM: the judgment gap is a symptom, not a separate problem
IBM’s Institute for Business Value surveyed 1,500 CHROs and 8,800 employees across 28 countries between April and June 2026 and found a split that reinforces the same story from the leadership side. Seventy-one percent of CHROs say supervising, validating and overriding AI outputs, in other words exercising judgment, is essential to how work should be done going forward. Only 29% of employees rank that same judgment as important to their own role. Sixty percent of employees say they worry AI is eroding their skills, and critical thinking is the skill they name most often as declining.
“AI is changing not only how work gets done, but where people can contribute the greatest value,” said Nickle LaMoreaux, senior vice president and chief human resources officer at IBM. That framing describes the intended destination. What the Visier and Gallup data show is that most organizations have not built the bridge to get there: they have handed employees a mandate to use AI without redesigning what judgment, contribution and value actually mean in the new arrangement. IBM’s study puts a second number next to the first that sharpens the point: 57% of CHROs cite critical thinking as the crucial workforce capability of the AI era, versus 49% of employees who rank it the same way. The two groups are not far apart on the answer. They are far apart on whether anyone has told employees the question is being asked of them at all.
Performative AI use is what fills that vacuum. It looks like adoption on a dashboard. It functions as a survival behavior.
Why the same gap keeps reopening
HR technology has run this pattern before, at smaller scale, with every prior wave of workplace software that promised to change how people work: enterprise collaboration suites, first-generation people-analytics platforms, remote-work tooling during the pandemic. In each case the rollout metric employers tracked first was adoption, licenses activated, logins per week, features touched, because adoption is the easiest thing to measure and report upward. What Gallup, Visier, Culture Amp and IBM are independently documenting this year is that AI has made the gap between adoption and understanding wider and more consequential than any of those earlier tools did, because AI output is harder to audit at a glance than a shared document or a dashboard, and because the stakes attached to it, in the form of the job-loss fears 24% of Visier’s respondents name outright, are higher. The same measurement habit that quietly under-served earlier rollouts is now producing a workforce that has learned to perform the metric rather than trust the explanation behind it.
What this means for the HR leader
The pattern across all four datasets suggests four practical failure points HR functions can address directly, rather than waiting on a strategy that will never feel finished enough to announce.
Stop measuring adoption alone
Usage metrics and license counts say nothing about whether people trust the guidance behind the tool. Visier’s finding that only 28% of leaders believe they understand actual employee AI use is a measurement failure before it is a communication failure. Pair adoption data with confidence and clarity questions in every pulse survey, the same way HRTech has reported employers doing in other contexts this year, or the dashboard will keep showing success while the workforce quietly hedges.
Say the plan is unfinished, out loud, on a schedule
Lavoie’s advice to trade certainty for candor is not a soft skill recommendation. It is a direct response to what happens when leaders go silent: employees fill the silence with self-protective behavior, including exaggerating their own fluency to avoid looking behind. A standing cadence, monthly or quarterly, where leadership states plainly what is known, what is not, and what is being tested closes more of the confidence gap than a single all-hands announcement ever will.
Redesign judgment into roles before asking people to exercise it
IBM’s judgment gap and Gallup’s finding that productivity gains concentrate on individual tasks rather than organizational redesign describe the same structural miss. If a CHRO expects employees to validate and override AI output, that expectation needs to be written into the role, trained for, and rewarded, not assumed. HRTech has previously documented how wide this specific gap runs between what CHROs expect and what employees are actually equipped to do, and the newer Visier and Gallup data suggests it has not closed since.
Do not roll out guidance the same way to every function
Gallup’s finding that leadership and technical roles report stronger AI benefits than service and administrative roles, and that healthcare and technical workers show the earliest adoption success, is a direct argument against a single company-wide AI memo. A workforce communication plan built for a technical team that already trusts the tool will read as empty reassurance to an administrative team still absorbing a 51% workflow change with no explanation of why. Segment the guidance the way the disruption is already segmented in the data.
None of the four organizations behind this data frame performative AI as a workforce integrity problem. Each frames it, in its own vocabulary, as a leadership communication gap. That is the more uncomfortable reading for HR functions that have spent 2026 treating adoption percentages as the scoreboard. The scoreboard was never the risk. The silence behind it was.
Source: Visier
